Skip to content

New: deep research, balance and auto-reload tools in the keyless MCP server. Install in one line

Response field

/v1/search and /v1/findSimilar (and their aliases) add one top-level field, contextfetch, that SDKs ignore. Read it from the raw HTTP response.

"contextfetch": {
  "cache": "miss",
  "timingsMs": {"recall": 727, "fetch": 953, "rerank": 2331, "total": 3488},
  "tokens": {"raw": 14200, "pruned": 1850, "savedPct": 87.0},
  "sources": {"searxng": 102, "stackoverflow": 10, "hn": 10, "github": 10, "wikipedia": 2, "local": 0}
}
FieldMeaning
cache"hit" when the same request was answered from the query cache, else "miss".
timingsMsServer-side milliseconds per stage and in total.
tokensMarkdown tokens of the returned results and of their prunedContext (ceil(chars / 4)); pruned is null unless highlights or pruned context were requested.
sourcesCandidates each recall source returned for this query.

The values in this example are illustrative, not measurements.